MSD 101: Fuzzy logic

Look-up tables are a convenient way to express the relationship between two variables in software. But the interpolation required between data points can overburden a microprocessor. Rather than sacrifice execution speed, some programmers turn to fuzzy logic.

The concept goes back to the 1960s, when UC-Berkeley professor Lotfi Zadeh developed “fuzzy set theory.” In contrast to standard set theory, where objects are either in or out of a set, fuzzy sets allow objects to have partial membership. Fuzzy logic advances the concept to include mathematical operations, most notably Boolean logic.

Questions & answers

Q What’s so special about fuzzy logic?
A It gives reasonable answers despite imprecise and conflicting data.

Q How should I select membership functions?
A Six or seven overlapping membership functions spanning the output range work best. Shape is relatively unimportant.

Q Should I use it instead of PID control?
A Not if you have a system that can be modeled as a second-order linear differential equation.